Leading Edge
Review
Integrating Biological Redesign:
Where Synthetic Biology Came From
and Where It Needs to Go
Jeffrey C. Way,1 James J. Collins,1,3 Jay D. Keasling,4,5,6,7 and Pamela A. Silver1,2,7,*
1Wyss
Institute for Biologically Inspired Engineering, Harvard University, Boston, MA 02115, USA
of Systems Biology, Harvard Medical School, Boston, MA 02115, USA
3Howard Hughes Medical Institute, Department of Biomedical Engineering and Center of Synthetic Biology, Boston University, Boston,
MA 02115, USA
4Department of Chemical and Biomolecular Engineering, Department of Bioengineering, University of California, Berkeley, Berkeley,
CA 94720, USA
5Physical Biosciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA
6Joint Bioenergy Institute, Emeryville, CA 94608, USA
7Synthetic Biology Engineering Research Center (SynBERC), University of California, Berkeley, Berkeley, CA 94720, USA
*Correspondence: [email protected]
http://dx.doi.org/10.1016/j.cell.2014.02.039
2Department
Synthetic biology seeks to extend approaches from engineering and computation to redesign of
biology, with goals such as generating new chemicals, improving human health, and addressing
environmental issues. Early on, several guiding principles of synthetic biology were articulated,
including design according to specification, separation of design from fabrication, use of standardized biological parts and organisms, and abstraction. We review the utility of these principles over
the past decade in light of the field’s accomplishments in building complex systems based on
microbial transcription and metabolism and describe the progress in mammalian cell engineering.
Introduction
Synthetic biology is a young discipline with the declared goal of
rationally engineering biological systems through approaches
similar to those used by engineers to build bridges and send
people to the moon. This field has rapidly developed over the
past 15 years from its initial conceptualization by a few academics and government program managers into a sizeable field
whose meetings attract large numbers of participants (http://
sb6.biobricks.org/).
The need for synthetic biology is certainly real. The moon-shot
challenges for the present generation of scientists working in the
field address a spectrum of urgent, real-world issues. It would be
ideal to engineer biofuel production from photosynthetic systems
to address the problems of global warming and energy self-sufficiency (Kung et al., 2012). Complex diseases such as cancer and
autoimmune disorders may simply never yield to simple target/inhibitor types of drug approaches and may require complex engineered proteins or cells (Bagshawe, 2009; Chester et al., 2002;
Porter et al., 2011). Food shortages may become an issue as
the world’s population continues to increase exponentially, while
production increases from the green revolution reach their limit
(Fischer et al., 2009; Bomford, 2009). In its move toward such
goals, synthetic biology also made an impact on the thinking
about biological systems, redefining organisms such as microbes previously appreciated for their basic biology or pathogenic properties, in terms of their value for biological redesign.
A framework for synthetic biology was proposed soon after its
inception as a discipline, and the developments in the last
decade provide an opportunity to look both backward and to
the future. This Review addresses where synthetic biology needs
to go to have a maximal impact and places it into the context of
existing disciplines.
Before Synthetic Biology: Recombinant DNA
The significance of the recombinant DNA revolution that started
in the mid-1970s cannot be overstated. It has led to new drugs,
metabolically engineered microbes that make diverse nutraceuticals and commodity biochemicals (e.g., most of the vitamins in vitamin pills), modern genotyping and DNA testing,
and genetic techniques that have transformed the way scientists
analyze biological systems. In the pharmaceutical industry
alone, recombinant proteins are used as monoclonal antibodies
and other protein drugs, as well as drug targets used in highthroughput screening. However, if we look at these accomplishments, it is important to note that recombinant DNA products
are limited in their complexity—single proteins, small molecules
based on screening against single proteins, and metabolically
engineered microbes that have been developed through extensive trial and error. We know how to make individual proteins,
but it goes beyond current practice to create new multicomponent systems that mirror natural biology in their complexity.
The promise of synthetic biology is to do just that. In particular,
the promise of synthetic biology is to construct organisms
with genuinely novel features that represent a jump from what
already exists. Such constructs can be thought of as working
systems that are separated from what currently exists by
Cell 157, March 27, 2014 ª2014 Elsevier Inc. 151
nonworking intermediates. Evolution generally requires working
intermediates. Because the systems that synthetic biologists
want to construct are complex and involve numerous variables
with values that must be in a narrow range for the system to
work (e.g., expression levels, binding constants, Hill coefficients, and geometry of interactions), it is often not feasible to
arrive at solutions simply by intuition, guesswork, or incremental
modification of an existing system. To think about an analogous
problem, consider the actual moon shot. This audacious program was not achieved through incremental changes from
existing systems, for example, by sending millions of randomly
varied test rockets to the moon, seeing which one makes it,
and then copying that design. The parallel to biology is that evolution can only get you so far over short time scales. Instead,
as with the moon shot, extensive design and calculation are
required to make big leaps to new biomolecular creations.
Formulating the Synthetic Biology Framework
Approaches to synthetic biology were most extensively explored
a decade ago, during a Defense Advanced Research Projects
Agency (DARPA)-sponsored study (Endy et al., 2003) that
brought together biologists, bioengineers, computer scientists,
and hardware developers and resulted in a recommendation
that the field of synthetic biology should learn from the earlier
developments in the microchip industry. This industry took dramatic steps forward in the late 1970s by setting up specific standards for chips, making them modular and easy to connect, and
through the formation of MOSIS, a DARPA-sponsored program
that supported the design and testing of chips (http://www.
mosis.com/). These measures rapidly led to the formation of
the integrated circuit industry, accompanied by several social
developments such as the separation of programming from
chip manufacturing. In synthetic biology, the corresponding
goal is to separate DNA synthesis from construct design and
to develop a class of bioengineers who would approach biology
with the mindset of computer programmers and chip designers.
The study ultimately led to the formulation of a synthetic biology
framework that, although not universally accepted, has defined
important elements in the thinking about how synthetic biology
could be practiced.
Design to Meet Specifications Set in Advance
This element posited that, as in other engineering fields, the
specifications for a design should be quantitative and relate to
the ultimate use of a genetic element. As an analogy, one can
think of a design of physically engineered elements—‘‘I need a
bridge that will support 50 tons of vehicles for 100 years’’ or ‘‘I
need a circuit board that will perform calculation X at a speed
of 106 calculations per second and operate between 0 C and
50 C.’’ Corresponding requirements for a biological system
might be ‘‘I need a genetic memory element that will maintain a
stable state with a spontaneous switching rate of <10 4 per
cell division in Gram-negative g-proteobacteria growing at a
rate of 0.3 to 3 doublings/hour.’’ Such specifications may ultimately become routine but currently are employed by only few
members of the synthetic biology community.
Separation of Design from Fabrication
In its most extreme form, this recommendation would correspond to a separation of synthetic biology projects into software
152 Cell 157, March 27, 2014 ª2014 Elsevier Inc.
and hardware efforts, with the software aspect focused on
design of constructs and the hardware aspect focused on making DNA, RNA, or proteins. The ideal outcome of this element
would be faster system testing.
Use of Biological ‘‘Parts’’
This principle relates to the issue of modularity in biology and
is intuitively obvious to most molecular biologists. The gene
has long been the central modular unit of biology, but with
the advent of molecular biology, genes, RNAs, and proteins
have been further dissected into functional modules such as
promoters and protein domains. An open question is whether
biology is genuinely modular in an engineering sense or whether
modularity is only a human construct that helps us understand biology. To clarify, it may be useful to distinguish between ‘‘concept modularity’’ and ‘‘engineering modularity.’’
For example, to understand the process of translation, it is useful to think of a ribosome binding site and a coding sequence
as separate modules. However, from an engineering point of
view, these elements are not necessarily distinct modules
because they can interact through mRNA secondary structure,
with levels of translation resulting from a nonlinear combination
of the two elements (Goodman et al., 2013; Gardner and Hawkins, 2013).
Abstraction of Biological Elements
Abstraction is an important aspect of how humans view and
engineer the world. In the computer industry, physical structures made of silicon, germanium, etc., are abstracted as
‘‘memory,’’ and a programmer can ignore what comprises the
memory. Biological engineers routinely perform such abstractions, representing proteins as circles, nucleic acids as lines,
and collections of metabolites as vectors in metabolic flux
models. Development of software tools for biology greatly
benefits from such abstraction; for example, the caDNAno program abstracts DNA structure to allow design of DNA origami
nanostructures (Douglas et al., 2009). One potential issue with
this recommendation is whether such abstractions interfere
with biological engineering by depicting as black boxes
elements that could and should be further characterized and
engineered.
Well-Defined ‘‘Chassis’’ Organisms
Another parallel to engineering is that a completely defined
cell would be ideal to serve as a chassis into which engineered
biological systems would be integrated. This approach could
potentially eliminate the vagaries associated with the unknown
features of a natural cell. The so-called ‘‘minimal cell’’ has
been sought through both top-down (removing as many genes
as possible; Hutchison et al., 1999; Sharma et al., 2007) and bottom-up approaches (resynthesis of new organisms, for example,
with alternative genetic codes, by combining the genome synthesis approach of Gibson et al. [2010] with the recoding strategy
of Lajoie et al. [2013]). A number of methods for genome editing
have surfaced in the last 10 years, making major changes in
genomic DNA faster and more predictable (Wang et al., 2009;
Gaj et al., 2013).
It has been 10 years since these ideas were put forth, and it
is thus worthwhile examining the success of the framework.
Not surprisingly, and as described in the following sections,
some elements have been more useful than others.
Employing Design Specifications: A Cultural Divide
Meeting design specifications in synthetic biology turned out
to be a major challenge because it often underscores the real difficulty of a given biological problem, especially for ambitious projects such as cell-based disease treatments. However, design
specifications are routinely used in metabolic engineering,
particularly with regard to production efficiency versus production cost. In our experience, the notion of design specifications
highlights a cultural divide between engineers and biologists.
Engineers learn the concept of design specifications in college,
whereas biologists trained in discovery research focus on uncovering new things about nature without preconceptions. Starting with a fixed idea of what is to be achieved, which is at the
core of synthetic biology, may inhibit discovery research while
being essential to the creation of new objects, whether physical
or biological. To enable synthetic biology, explicit training in
design-to-specification will be needed at an early stage in the
careers of biologists.
Separating Design and Fabrication: Hardware Outpaces
Software
The implementation of this element has been decidedly
mixed. On the hardware side, DNA synthesis is becoming
cheaper and cheaper. Following the microchip analogy,
Carlson (2009) described this phenomenon as exceeding
‘‘Moore’s Law,’’ which, describing the rapid rise of integrated circuits, postulated that increases in computer power double approximately every 2 years. On the other
hand, although there is an ongoing evolution toward
outsourcing DNA constructions, graduate students and
postdocs still spend much of their time making plasmids.
The real transition will come when DNA synthesis is so
cheap that a graduate student no longer needs to ask an
advisor for permission to build a new microbial genome
or an entire plant chromosome, for example. The design
aspect has been more challenging. The Platonic ideal
would be ordering a set of DNA constructs and placing
them in an organism where they would work with minimal
tweaking and optimization. What is necessary for such a
successful design? There has been some effort in the
development of design tools that address very specific purposes, such as flux-balance programs for metabolism,
translation calculators for optimization of ribosome binding
sites (Salis, 2011), caDNAno for designing oligonucleotidebased origami structures (Douglas et al., 2009), protein
structure visualization programs, RosettaDesign for engineering protein domains (Tinberg et al., 2013), as well as
tools for engineering multidomain proteins (RobinsonMosher et al., 2013), and numerous other pieces of software (Kahl and Endy, 2013). Many applications in synthetic
biology would require most or all of these tools—in the
course of evolution, nature changes all of these features
at the same time. At present, the majority of these computer-based design tools allow ruling out designs with fatal
flaws, but extensive trial and error is often still required to
obtain a working prototype. Moreover, the use of these
tools requires both a familiarity with the tools themselves
and a decent understanding of the biological process being
modulated; such a breadth of understanding often goes
beyond current educational practice.
Biological Modules: Learning from Nature
The use of biological modules has been championed by the Registry of Biological Parts, SynBERC, and others (Shetty et al.,
2008; Kosuri et al., 2013; Kobayashi et al., 2004; Mutalik et al.,
2013a, 2013b). The underlying challenge is to collect, characterize, and catalog DNA elements with potential utility. A registry
serves as a collection point for information, as well as for DNA
elements themselves, although the latter function may become
irrelevant as DNA synthesis becomes sufficiently cheap to allow
creation of such ‘‘parts’’ from scratch, rather than obtaining them
from a registry and assembling them. A registry would then
remain a storage point for information and standards. Another
element of the original concept was devising a universal connecting method to facilitate construction of progressively more
complex parts. The original BioBricks standard (Knight, 2003)
involves a restriction enzyme-based cloning strategy in which
ligation eliminates the medial cut-sites so that the resulting
DNA becomes a larger part. This approach is becoming less
relevant in light of new technology; specifically, the assembly
method of Gibson et al. (2009) and other related strategies
(Kosuri et al., 2010; Dharmadi et al., 2013), which are restriction-site-independent methods for assembling DNA and which
may someday replace most restriction enzyme-based cloning.
The design of artificial modules has been often inspired by natural genetic modules, such as the lactose operon and pathogenicity islands in bacteria. It is, however, important to point out
that the modularity of these units is a consequence of natural
selection—they are thought to be transferred between different
types of bacteria by transduction and conjugation, becoming
self-sufficient through natural selection, and to able to function
in diverse bacteria that may vary in other ways, such as their
central metabolism or ribosomal RNAs that recognize ShineDalgarno sequences. Other biological processes, such as cell
membrane construction or chromosomal replication, may be
nonmodular, but this will be less apparent from the genetic organization. Protein domains constitute another example of genetic
modules. There are numerous examples of large proteins that
have presumably arisen from fusion of smaller units, and it
appears that such fusions are a common occurrence during
evolution. For example, the SH2, SH3, protein kinase, fibronectin
Type III, immunoglobulin, and WD40 domains are used in
signaling proteins in a wide variety of configurations and contexts, arising by exon shuffling and similar mechanisms (Gilbert,
1978; Bork et al., 1997). It seems likely that, after initial formation,
new fusion proteins may be suboptimal and may further evolve
by point mutations and changes in interdomain segments to
adjust to their biological context. The point for synthetic biologists is that modularity can be a useful place to start, but additional conceptual tools are needed to improve protein constructs
before and after modules are assembled (Cironi et al., 2008;
Robinson-Mosher et al., 2013).
Abstraction: A Double-Edged Sword
In the context of biological engineering, it is still an open question whether abstraction is a useful tool or a necessary evil.
Cell 157, March 27, 2014 ª2014 Elsevier Inc. 153
Figure 1. Synthetic Biological
Based on Gene Expression
Devices
(A) The toggle switch, in which two repressors turn
each other off, leading to a bistable transcriptional
state (Gardner et al., 2000).
(B) The trigger-memory system, in which an environmental stimulus induces an activator (the
trigger) that then turns on its own synthesis in a
second ‘‘memory’’ element, leading to a permanent ‘‘ON’’ state (Ajo-Franklin et al., 2007; Burrill
and Silver, 2011; Burrill et al., 2012).
(C) A DNA-based memory element, in which activation of an integrase or integrase + excisionase
leads to alternating DNA states (Bonnet et al.,
2012, 2013; Siuti et al., 2013).
(D) The repressilator, in which three repressors
sequentially turn each other off, such that the state
of the system oscillates in time (Elowitz and Leibler,
2000).
(E) An event counter, in which a signal turns on
a positive regulator of translation and acts in
combination with sequentially expressed transcriptional activators to count pulses of a signal
(Friedland et al., 2009).
154 Cell 157, March 27, 2014 ª2014 Elsevier Inc.
Computer hardware can be abstracted to the point where a circuit can be represented simply in terms of its inputs and outputs.
Our ability to perform such an abstraction results from our ability
to design computer hardware around human cognitive capabilities. Biological systems, in contrast, have been created by evolution and are not necessarily abstractable in ways that the
human brain has evolved to handle. For example, the massively
parallel function of metabolism, Brownian motion, allosteric
changes in protein conformation, and movement in a low Reynolds number regime are examples of processes that are difficult
to abstract and intuit, which may lead to poor design in biological
engineering. The challenges posed by abstraction can be illustrated by two examples from the synthetic biology field.
The ‘‘Bag of Genes’’ Abstraction of a Cell
Synthetic biologists have spent a great deal of time developing
artificial transcriptional systems (Figure 1). In fact, one of us
(J.J.C.) recently edited an issue of the journal Chaos focused
on synthetic biology, with most articles describing transcriptional systems (Albert et al., 2013). Transcriptional networks
are attractive because their behavior can be modeled as a
‘‘bag of genes’’—analogous to the biochemists’ ‘‘bag of enzymes’’—through various reasonable abstractions and because
the key biochemical values have been determined for a number
of bacterial systems (Huynen and Bork, 1998; Mathews, 1993;
Arkin et al., 1998). For example, the bacteriophage l system
has been a favorite of mathematical modelers, who have rationalized the lysis/lysogeny decision and the decision to exit or
enter the lysogenic state (Ackers et al., 1982; Arkin et al.,
1998). The modeling is possible because many of the quantitative parameters of these systems have been determined over
the past 40 years. Based on this background, Gardner et al.
(2000) and Elowitz and Leibler (2000) constructed simple transcriptional devices that behaved either as a toggle switch or an
oscillator (Figures 1A and 1D). These early examples of synthetic-biological circuits suggested the possibility of designing
artificial transcriptional circuits that could perform useful tasks.
Since then, there has been much follow-up work on transcriptional circuits. Kobayashi et al. (2004) demonstrated that it is
possible to link promoters to output molecules to construct
new genetic devices and programmable cells, which was a natural extension of the use of constructs for reporters, protein
expression, and protein relocalization that had been used since
the mid-1970s (Casadaban, 1976; Backman and Ptashne, 1978;
Silhavy et al., 1976). Use of RNA binding to an mRNA offers
another regulatory mode that is easily programmed, building
on base-pairing rules and energetic considerations that can be
translated into simple design rules (Isaacs et al., 2004; Callura
et al., 2010). Certain DNA binding proteins such as zinc-finger
(ZF) proteins, and especially TALEs, have a roughly one-to-one
correspondence between certain amino acids and the DNA
that is recognized, allowing straightforward programming of
transcription as well (Elrod-Erickson et al., 1998; Garg et al.,
2012; Lienert et al., 2013; Bultmann et al., 2012; Khalil et al.,
2012). These features make plausible the idea that biological
systems could be programmed by simple rules to allow biological computation and have thus attracted a great deal of attention from the synthetic biology community. That being said, the
focus on gene expression has its limitations. First, gene-expres-
sion-based circuits are relatively slow. Even in bacteria, transcription plus translation to yield an average-sized protein (340
aa at 17 aa/s translation; Young and Bremer, 1976) would take
about 20 s, compared to the subnanosecond calculation speed
in a typical personal computer. Biological systems do perform
rapid computations, using neurons for example, but these fast
mechanisms do not involve transcription or translation. Second,
the inputs and outputs will likely be the more interesting things
about a biological circuit. DNA-protein interactions may be
somewhat programmable, but interactions with the chemical
and physical environment are much more difficult to engineer.
A holy grail of synthetic biology and protein engineering is to
design macromolecules that can recognize an arbitrarily chosen
ligand, with coupling to an arbitrary enzymatic activity. Artificial
fusion proteins with allosteric properties have been constructed
(Guntas et al., 2005; Meister and Joshi, 2013), but their function
may depend on the particular geometry of the fusion partners,
and it is difficult to know how general this approach may be.
RNAs have been identified that bind to various ligands, and these
have been further engineered to create sensors that regulate
gene expression (Win et al., 2009). However, it would be ideal
to be able to start with the large number of natural ligand-binding
elements, most of which are protein based, rather than needing
to start a project by isolating RNA aptamers that bind to a target
of interest. This approach has led to biological computers of
general programmability that can be built from transcriptional
systems. These are slow and will never approach the speed of
silicon-based computers. Their real value will likely be that biological computers can be embedded in the environment and
‘‘live off the land’’ while recording and calculating events of interest. Genetic memory devices such as the toggle switch and its
offspring could be particularly useful in this regard (Gardner
et al., 2000; Ajo-Franklin et al., 2007; Burrill and Silver, 2011;
Burrill et al., 2012; Bonnet et al., 2012, 2013).
The ‘‘Bag of Enzymes’’ Abstraction of a Cell and Its
Limitations
Metabolic engineering involves the genetic manipulation of
microbes to overproduce compounds of commercial interest
(Keasling, 2012; Keasling et al., 2012). Synthetic biology has
the potential to significantly extend the reach of metabolic engineering through the use of fused and scaffolded enzymes to
direct flux and enhance yields (Dueber et al., 2009; Agapakis
et al., 2010; Delebecque et al., 2011; Whitaker and Dueber,
2011) and through the creation of novel pathways that can synthesize natural or nonnatural compounds (Moon et al., 2009;
Steen et al., 2010). If properly harnessed, enzymatic chemistry
could create thousands of diverse, novel molecules and materials. The main tools of metabolic engineering have been gene
knockouts, addition of foreign pathways into high-producing
hosts, and analysis of flux patterns. One programmable aspect
of metabolic engineering derives from simple conservation of
mass and has led to genome-scale flux-balance models of
cellular metabolism (Feist and Palsson, 2008; Oberhardt et al.,
2009). This approach is the essence of the ‘‘bag of enzymes’’
view of the cell—the cytoplasm is a vessel in which enzymes
and substrates are evenly distributed throughout the cell.
Recently, several groups have engineered channeling and the
compartmentalization of metabolism with the goal of enhancing
Cell 157, March 27, 2014 ª2014 Elsevier Inc. 155
be that there are no regulatory proteins that can sense such
novel end products, so engineering feedback inhibition may be
difficult (Figure 2), underscoring the need for the integration of
protein engineering into synthetic-biological systems.
Figure 2. Control of Flux in a Cell Making a Nonnatural Product
In natural metabolism, almost every branch-point step is transcriptionally and/
or allosterically regulated. One goal of synthetic biology is to make nonnatural
products, which may be achieved by piecing together an artificial pathway of
enzymes that may be from different sources or engineered to have new
specificities. However, in such cases, there may be no mechanism for regulating the branch point at which flux is siphoned off from central metabolism.
This problem is particularly acute if some carbon flux through central metabolism is needed to generate ATP or reducing equivalents. It would be ideal to
engineer compound recognition into either a transcription factor or an allosteric site, but this is beyond our current protein engineering capability.
overall rates and limiting side reactions. Cells naturally use compartments to protect the contents from the cytoplasm, as well as
the cytoplasm from the contents, to create a unique chemical
environment and to concentrate chemical reactions (Mathews,
1993; Cheng et al., 2008). Dueber et al. (2009) placed three enzymes catalyzing mevalonate synthesis on a protein scaffold
and observed an almost 80-fold synthesis enhancement. Delebecque et al. (2011) placed hydrogen-synthesizing proteins on
an RNA scaffold and found a progressive increase of product
synthesis as the dimensionality of the scaffold was increased
from a zero-dimensional single RNA, a linear scaffold, to a
2D scaffold that gave an 40-fold increase. These approaches
draw on natural examples of substrate channeling and illustrate
how moving beyond a simplistic conception of the cell as a ‘‘bag
of enzymes’’ can be useful (Mathews, 1993). The synthesis of
compounds that are not normally made by biological systems
is an important new area at the intersection of synthetic biology
and metabolic engineering. The goal is to use biological chemistry to make molecules that might otherwise be made using synthetic chemistry. For example, Atsumi et al. (2008) produced a
set of related alcohols, some known and some new to biology,
by mixing and matching enzymes to drive the production
of various 2-keto acids, followed by decarboxylation with a
nonspecific 2-keto acid decarboxylase and an alcohol dehydrogenase to reduce the resulting aldehyde. Similarly, Steen et al.
(2010) described the synthesis of a variety of biofuel-type
molecules and precursors, including C12- and C14-length fatty
alcohols that are not normally produced biologically but whose
production is engineered by coexpressing medium-chain fatty
acyl thioesterases and nonspecific reductases from different
organisms. Moon et al. (2009) created an artificial pathway for
production of glucaric acid by combining enzymes from E. coli,
yeast, and mammals. These examples illustrate how combining
enzymes from diverse sources into a single cell can yield novel
chemicals that are normally made only by synthetic organic
chemistry or are not made at all. An important next step will likely
be to combine this approach with protein engineering to further
diversify the space of biochemicals. One future challenge may
156 Cell 157, March 27, 2014 ª2014 Elsevier Inc.
Chassis Organisms: Moving beyond Microbes
An ideal chassis organism would be one that is so well understood that, when engineered DNA is added, there are no surprise
interactions between the added material and host functions. One
scenario would be a ‘‘minimal’’ organism—an organism built
from the bottom up in which each gene and protein is present
for a reason. Some effort has gone into generating such organisms by systematic gene deletion in an organism with an already
small genome, for example (Glass et al., 2006). However, for microbial engineering, it is not clear whether unexpected or fortuitous interactions between added and endogenous factors might
be a problem (though minimal organisms may help us test it).
This is in sharp contrast to mammalian cell engineering in which
one major problem is that unpredictable long-range interactions
along the DNA in mammalian chromosomes occur all the time.
Much of the preceding discussion applies to microbial synthetic biology but does not carry over to mammalian cells. In
particular, the bedrock technologies of synthetic biology—complex but reproducible genetic engineering and quantitative control—are not currently as feasible in mammalian cell engineering.
A mammalian chassis for synthetic-biological circuits would
have great utility.
Mammalian cell engineering, which could have a profound
impact on treatment of disease, can boast of several major accomplishments. Production of monoclonal antibodies and other
proteins has had a profound impact on the pharmaceutical industry, with protein drugs comprising almost 50% of total drug
revenues and a higher efficiency of approval than chemical drugs
(Rader, 2012). A few of these proteins are made in bacteria or
yeast, but most are produced from mammalian cells because
these cells mediate natural mammalian glycosylation, are better
at catalyzing the folding of many proteins, and do not produce
inflammatory contaminants (e.g., bacterial cell wall material)
that must be completely removed before injection. However,
the engineering of mammalian cells often involves laborious,
inefficient methods. Cells are transfected with expression constructs that are typically integrated randomly into the genome
(Murnane et al., 1990; Merrihew et al., 1996), although new
targeted integration technologies may change this (Gaj et al.,
2013). Expression levels vary depending on the integration site,
as well as on epigenetically controlled chromatin structure
around the inserted transgene. It is possible to identify cell
clones that constitutively express one or two proteins (e.g., antibody-heavy and light chains) by brute-force screening. This is
adequate for the pharmaceutical industry because the cost of
this method is small compared to that of a clinical trial. However,
more complex systems will require faster, more reproducible
technologies for engineering mammalian cells.
There has been some success in engineering genetic circuits
in mammalian cells. As with bacteria, much of the effort has
focused on gene expression-based devices (Deans et al.,
2007). Transcriptional logic-based circuits analogous to those
constructed in bacteria have also been engineered into
Figure 3. Biological Understanding along
the ‘‘Describe-Explain-Predict-Control’’ Axis
The figure diagrams a rough estimate of how well
we understand and can engineer various biomolecular elements and events. Some, such as
nucleic acid interactions with other nucleic acids
and certain proteins, are so well understood that
they can be predictably engineered with minimal
postbuild testing. Others, such as movement of
proteins during allosteric transitions or along
cytoskeletal elements, can be described and their
behavior sometimes rationalized post hoc, but to
engineer such systems, it is often difficult even to
know where to start. Fully controlling the remarkable capabilities of cells and organisms will require
integrated engineering of the extensive phenomena of biology.
mammalian cells, but there are also challenges in construct assembly still being addressed (Wieland and Fussenegger, 2012;
Wei et al., 2013; Guye et al., 2013). For example, Burrill et al.
(2012) constructed memory systems in mammalian cells, in
which trigger and memory elements were sequentially transfected into cells. A single trigger element construct was inserted
randomly into the genome by standard procedures, and clones
were screened for the ability to respond to an inducing signal
(doxycycline, DNA damage, and hypoxia); clones with the correct behavior were then transfected with a memory element
and screened for the ability to retain a memory of the inducing
signal. In this approach, quantitative adjustment of the promoter
and ribosome-binding site (RBS) strength was replaced by varying the insertion site and letting the influence of local chromatin
structure determine the level of gene expression. This approach
does not fulfill the desire for precise and predictable engineering
of mammalian cells. Recently, new tools have emerged that
should allow specific integration at desired sites in the genome.
For example, methods based on zinc-finger, TALE, and CRISPR
fusions to nucleases can be used to generate double-strand
breaks at specific sites in the genome (Gaj et al., 2013). The
questions remain—where should we integrate, and how can
we avoid effects of adjacent sequences?
What is needed to make mammalian cells engineerable? Scientific understanding can be oriented along an axis of ‘‘describeexplain-predict-control’’ (Evans and Rooney, 2008; Figure 3).
Elements of bacterial gene expression lie toward the ‘‘control’’
side of this continuum, but gene expression in multicellular organisms is still somewhere in the middle. A variety of elements
that are important in metazoan gene expression have been identified, including enhancers, TATA elements, silencers, and insulators, but it is not clear how to combine these elements to create
new regulatory systems with quantitatively predictable properties. Many of these elements are understood in some mechanistic detail, but this information is distributed throughout the
scientific literature, making it difficult to extract the key bits of
information that are useful for engineering. A particular issue
that is somewhat unique to mammalian cells is that long-range
genomic context effects have a profound effect on expression
of transgenes (Carroll et al., 2005; Ribeiro de Almeida et al.,
2012). It is difficult to simply look (even with computers) at a
genome sequence and figure out where the enhancers, insula-
tors, and sometimes even coding sequences are located. In
addition, the long-range nature of metazoan transcriptional control means that even transgenes may interfere with each other.
For effective, predictable engineering of mammalian cells, it
may be necessary to construct a well-understood artificial
chassis chromosome. Such an element would need to have
defined insertion sites, insulators that functionally separate
transgenes from each other, and physical separation from the
native genome to avoid unpredictable long-range effects. There
has been some effort to construct artificial chromosomes ‘‘topdown’’ by deletion of material from an existing chromosome
(Ren et al., 2006; Bergmann et al., 2012), but remaining
undeleted material may still influence expression of inserted
transgenes. Piecing together all of the elements of an artificial
chromosome by a ‘‘bottom-up’’ approach is possible in principle
but would require that we know all of the pieces. Hopefully, one
or both of these approaches will bear fruit, because the need for
a mammalian chassis chromosome is critical.
Mammalian signal transduction can also be engineered. Yeh
et al. (2007) modified a signal transduction pathway that controls
the actin cytoskeleton to alter cell shape in response to an
external signal, a response that did not involve gene expression.
Several groups have also engineered T cells to recognize tumor
cells by expression of chimeric antigen receptors (CARs; Gross
et al., 1989; Porter et al., 2011). These elements consist of single-chain antibody V regions fused to the z chain of the T cell receptor; the V regions recognize tumor-specific surface antigens.
The CAR construction is a synthetic-biological creation with an
obvious practical utility. As such, it evokes the words of Steve
Jobs (Feld, 2011): ‘‘You’ve got to start with the customer experience and work backwards to the technology. You can’t start
with the technology and try to figure out where you’re going to
try to sell it.’’ The creators of CARs started with a particular
‘‘customer experience’’ goal, treating cancer, and then developed a technology for doing so. Development of CARs did not
appear to begin with design specifications beyond wanting to
kill cancer cells; such specifications might have included a
therapeutic index that could be optimized by adjusting the binding constant of the antibody for its targets. Nonetheless, the
approach works at some level in leukemias in which the target
cells are accessible (Porter et al., 2011), and quantitative optimization may help when the antigen is not completely tumor
Cell 157, March 27, 2014 ª2014 Elsevier Inc. 157
specific or for solid tumors in which the cells are less accessible.
The CAR example illustrates the value of engineering signaling
proteins that are far removed from transcription, even though
we do not have particularly elegant tools for doing so.
Designing to Meet Biological Grand Challenges
In sum, the original guiding principles conceptualized for synthetic biology a decade ago have permeated a widespread,
diverse community, but it remains to be seen which elements
merit adoption as research becomes more translational. For
this to happen, it is worth considering synthetic biology
approaches in light of the challenges posed at the beginning of
this Review: designing microbes that produce biofuels that can
go directly into a gas tank; developing immune cells that kill metastatic solid tumor cells; and creating engineered food crops
with increased per acre yield and reduced fertilizer and water
use. Can synthetic biology deliver these products? Following
Steve Jobs, let us look ahead to the final product that we want
and then work backward to the design specifications.
For living cells to make biofuels, the specifications might be
that the fuel molecules would spontaneously separate from the
aqueous phase and from cell membranes, would not be toxic
to the cells, would have appropriate combustion properties,
and would be efficiently produced, etc. To have appropriate
combustion properties, it is likely necessary to synthesize a
mixture of molecules similar to fuels from petroleum feedstocks,
such as a mix of linear and branched medium-length alkanes,
isoprenoids, and related molecules. To efficiently synthesize
the medium-length linear alkanes, one could modify fatty acid
synthesis such that it terminates before the usual C16- to C18length molecules and express enzymes to make alkanes or fatty
alcohols (Schirmer et al., 2010; Steen et al., 2010) but with
shorter-than-usual length specificity, so extensive protein engineering would be required. In addition, the producing cell would
need to make normal fatty acids to grow and then switch on production of the alternative pathway, while switching off synthesis
of full-length fatty acids and possibly cell growth itself to minimize the waste of carbon feedstock. To insure correct distribution of acetyl-CoA into fatty acid initiation and elongation, it
may be necessary to adjust the natural mechanism to reflect
the shorter chain length of the desired product. Coexpression
of certain amino acid and isoprenoid pathways would yield a
useful mix of branched carbon chains. Thus, the final engineered
organism would require both extensive protein and transcriptional engineering and, possibly, engineering of the novel
enzymes into a scaffolded structure to enhance synthesis of
the specific product. All of these engineering modes would
need to be tightly interdigitated.
The design specifications for engineered immune cells that
successfully attack solid tumors would be that enough metastatic tumor cells are killed to cause a complete remission possibly
with lasting immunity, that the side effects are tolerable, and that
systemic autoimmunity does not result. The challenges for any
immunological therapy to treat solid tumors are formidable.
Unlike leukemias and lymphomas, solid tumor cells will be
separated from the engineered T cells by a barrier of capillary
endothelial cells and a lack of lymph node drainage to draw anything into the tumor. The T cells might first need to be engineered
158 Cell 157, March 27, 2014 ª2014 Elsevier Inc.
to recognize the tumor-specific endothelium with coupling to
the system that mediates tissue invasion. Once in the tumor,
the T cells would specifically recognize and kill the tumor cells,
in spite of the numerous immunosuppressive cytokines being
secreted by the tumor. The number of cells killed per attacking
T cells must likely be higher than normal, so the mechanisms
for downregulating T cell killing would need to be modified, but
not in a way that would allow attack on normal cells. Finally,
the engineered cells would need to die or be easily killed after
the tumor is eliminated. Such engineering could build on the
CAR-expressing T cells but would be several times more complicated. The behavior of solid tumors in humans is different than in
mouse models because the human tumors grow more slowly,
and some underlying immunology is simply different between
the two organisms (Mestas and Hughes, 2004). Therefore the
design may need to be particularly theory driven, with limited
experimental support before a Phase 1 clinical trial. This truly
would resemble a moon shot—the calculations would need to
be right.
The increases in the per-acre yields of food crops have been
flattening over the past 15 years or so, after dramatic increases
of up to 6-fold that occurred between the 1960s and 1990s
(Fischer et al., 2009). The S-shaped curve for crop yields contrasts with the exponential growth of the human population,
and new sources of food will be needed. The green revolution
combined traditional crop breeding with intensified water and
fertilizer use, which led to increased crop yields but with an
environmentally unsustainable approach, particularly in its use
of nitrogeous fertilizers. One design specification would be that
plants simply fix their own nitrogen. This would presumably
involve introduction of a large number of genes from N2-fixing
bacteria, plus regulation and compartmentalization so that the
genes are expressed in an oxygen-free environment.
Conclusions
Synthetic biology is still in its early stages of development. If we
stick with the comparison to the microchip industry and consider
that the first transistor was developed in 1947, then we are
now at about 1960. In the near future, we also hope that the
approaches of synthetic biology will have permeated the biological community. In particular, the most important engineering
ideas—design to specification and extensive pre-experiment
planning to avoid excessive trial and error—need to be adopted
because it should be more cost effective than the current style.
For synthetic biology to go beyond a self-selected community,
it will need successes with clear-cut utility that clearly derive
from the field’s specific approach.
If synthetic biologists are to harness the power of biology, they
will need to have a command of all the things that biology modulates during the course of evolution—e.g., gene expression,
metabolism, protein structure, and function. Nature manipulates
all of these things at the same time, and synthetic biologists need
to be able to do so as well. The redesign of biology will necessarily involve an integration multifaceted engineering skills.
This leads to a key question for the future development of synthetic biology: do human engineers need to understand biology
as it exists with all of its nonintuitive quirkiness, or can biology be
redesigned so that engineers work with a simplified form that is
adapted to the creative processes of the human brain? In the
case of computers, the latter is the case—materials scientists
have adapted inorganic matter to be used in transistors in a
way that is physically robust and also abstractable. Unfortunately, as of now, the approaches of synthetic biology do not
always allow for a single solution that can be used over and
over again. Moreover, some accomplishments of nature are
hard for humans to reproduce—for example, protein engineers
have not yet built de-novo-designed proteins, particularly enzymes, that are truly useful in biological engineering. It therefore
seems likely that most synthetic biology will involve repeated,
intensive borrowing from nature, and the most effective practitioners in the further development of the field will be those with
an engineering mindset who, functioning alone or in integrated
teams, understand most broadly how natural biology works.
ACKNOWLEDGMENTS
We acknowledge the Wyss Institute for continued support, the original
Synthetic Biology Working Group at MIT, the many participants in iGEM since
its inception, past and present members of our laboratories, a few visionary
government scientists and program managers from DARPA and ONR who
appreciated the potential of the field before it existed, and Devin Burrill
for comments on the manuscript. The authors acknowledge support over
the years from DARPA, ONR, HHMI, DOE, NIH, NSF, and their respective
institutions.
REFERENCES
Ackers, G.K., Johnson, A.D., and Shea, M.A. (1982). Quantitative model for
gene regulation by lambda phage repressor. Proc. Natl. Acad. Sci. USA 79,
1129–1133.
Agapakis, C.M., Ducat, D.C., Boyle, P.M., Wintermute, E.H., Way, J.C., and
Silver, P.A. (2010). Insulation of a synthetic hydrogen metabolism circuit in
bacteria. J. Biol. Eng. 4, 3.
Ajo-Franklin, C.M., Drubin, D.A., Eskin, J.A., Gee, E.P., Landgraf, D., Phillips, I.,
and Silver, P.A. (2007). Rational design of memory in eukaryotic cells. Genes
Dev. 21, 2271–2276.
Albert, R., Collins, J.J., and Glass, L. (2013). Introduction to focus issue: quantitative approaches to genetic networks. Chaos 23, 025001.
Arkin, A., Ross, J., and McAdams, H.H. (1998). Stochastic kinetic analysis of
developmental pathway bifurcation in phage lambda-infected Escherichia
coli cells. Genetics 149, 1633–1648.
Atsumi, S., Hanai, T., and Liao, J.C. (2008). Non-fermentative pathways for
synthesis of branched-chain higher alcohols as biofuels. Nature 451, 86–89.
Bultmann, S., Morbitzer, R., Schmidt, C.S., Thanisch, K., Spada, F., Elsaesser,
J., Lahaye, T., and Leonhardt, H. (2012). Targeted transcriptional activation of
silent oct4 pluripotency gene by combining designer TALEs and inhibition of
epigenetic modifiers. Nucleic Acids Res. 40, 5368–5377.
Burrill, D.R., and Silver, P.A. (2011). Synthetic circuit identifies subpopulations
with sustained memory of DNA damage. Genes Dev. 25, 434–439.
Burrill, D.R., Inniss, M.C., Boyle, P.M., and Silver, P.A. (2012). Synthetic memory circuits for tracking human cell fate. Genes Dev. 26, 1486–1497.
Callura, J.M., Dwyer, D.J., Isaacs, F.J., Cantor, C.R., and Collins, J.J. (2010).
Tracking, tuning, and terminating microbial physiology using synthetic riboregulators. Proc. Natl. Acad. Sci. USA 107, 15898–15903.
Carlson, R. (2009). The changing economics of DNA synthesis. Nat. Biotechnol. 27, 1091–1094.
Carroll, J.S., Liu, X.S., Brodsky, A.S., Li, W., Meyer, C.A., Szary, A.J.,
Eeckhoute, J., Shao, W., Hestermann, E.V., Geistlinger, T.R., et al. (2005).
Chromosome-wide mapping of estrogen receptor binding reveals long-range
regulation requiring the forkhead protein FoxA1. Cell 122, 33–43.
Casadaban, M.J. (1976). Transposition and fusion of the lac genes to selected
promoters in Escherichia coli using bacteriophage lambda and Mu. J. Mol.
Biol. 104, 541–555.
Cheng, S., Liu, Y., Crowley, C.S., Yeates, T.O., and Bobik, T.A. (2008).
Bacterial microcompartments: their properties and paradoxes. Bioessays
30, 1084–1095.
Chester, J., Ruchatz, A., Gough, M., Crittenden, M., Chong, H., Cosset, F.L.,
Diaz, R.M., Harrington, K., Alvarez-Vallina, L., and Vile, R. (2002). Tumor antigen-specific induction of transcriptionally targeted retroviral vectors from
chimeric immune receptor-modified T cells. Nat. Biotechnol. 20, 256–263.
Cironi, P., Swinburne, I.A., and Silver, P.A. (2008). Enhancement of cell type
specificity by quantitative modulation of a chimeric ligand. J. Biol. Chem.
283, 8469–8476.
Deans, T.L., Cantor, C.R., and Collins, J.J. (2007). A tunable genetic switch
based on RNAi and repressor proteins for regulating gene expression in
mammalian cells. Cell 130, 363–372.
Delebecque, C.J., Lindner, A.B., Silver, P.A., and Aldaye, F.A. (2011). Organization of intracellular reactions with rationally designed RNA assemblies.
Science 333, 470–474.
Dharmadi, Y., Patel, K., Shapland, E., Hollis, D., Slaby, T., Klinkner, N., Dean,
J., and Chandran, S.S. (2013). High-throughput, cost-effective verification of
structural DNA assembly. Nucleic Acids Res.. http://dx.doi.org/10.1093/nar/
gkt1088.
Douglas, S.M., Marblestone, A.H., Teerapittayanon, S., Vazquez, A., Church,
G.M., and Shih, W.M. (2009). Rapid prototyping of 3D DNA-origami shapes
with caDNAno. Nucleic Acids Res. 37, 5001–5006.
Backman, K., and Ptashne, M. (1978). Maximizing gene expression on a
plasmid using recombination in vitro. Cell 13, 65–71.
Dueber, J.E., Wu, G.C., Malmirchegini, G.R., Moon, T.S., Petzold, C.J., Ullal,
A.V., Prather, K.L., and Keasling, J.D. (2009). Synthetic protein scaffolds
provide modular control over metabolic flux. Nat. Biotechnol. 27, 753–759.
Bagshawe, K.D. (2009). Targeting: the ADEPT story so far. Curr. Drug Targets
10, 152–157.
Elowitz, M.B., and Leibler, S. (2000). A synthetic oscillatory network of transcriptional regulators. Nature 403, 335–338.
Bergmann, J.H., Martins, N.M., Larionov, V., Masumoto, H., and Earnshaw,
W.C. (2012). HACking the centromere chromatin code: insights from human
artificial chromosomes. Chromosome Res. 20, 505–519.
Bomford, P. (2009). Crop yield projections for biofuels fall short. Post
Carbon Institute, September 4, 2009. http://www.postcarbon.org/article/
40439-crop-yield-projections-for-biofuels-fall.
Bonnet, J., Subsoontorn, P., and Endy, D. (2012). Rewritable digital data
storage in live cells via engineered control of recombination directionality.
Proc. Natl. Acad. Sci. USA 109, 8884–8889.
Elrod-Erickson, M., Benson, T.E., and Pabo, C.O. (1998). High-resolution
structures of variant Zif268-DNA complexes: implications for understanding
zinc finger-DNA recognition. Structure 6, 451–464.
Endy, D., Lincoln, P., and Murray, R. (2003). 2003 Synthetic Biology Study.
http://hdl.handle.net/1721.1/38455.
Evans, A.N., and Rooney, B.J. (2008). Methods in Psychological Research
(Thousand Oaks, CA: Sage Publications).
Bonnet, J., Yin, P., Ortiz, M.E., Subsoontorn, P., and Endy, D. (2013). Amplifying genetic logic gates. Science 340, 599–603.
Feist, A.M., and Palsson, B.O. (2008). The growing scope of applications
of genome-scale metabolic reconstructions using Escherichia coli. Nat.
Biotechnol. 26, 659–667.
Bork, P., Schultz, J., and Ponting, C.P. (1997). Cytoplasmic signalling
domains: the next generation. Trends Biochem. Sci. 22, 296–298.
Feld, B. (2011) Start with customer experience. http://www.feld.com/wp/
archives/2011/09/start-with-customer-experience.html.
Cell 157, March 27, 2014 ª2014 Elsevier Inc. 159
Fischer, R.A., Byerlee, E., and Edmeades, E.O. (2009). Can technology deliver
on the yield challenge to 2050? ftp://ftp.fao.org/docrep/fao/012/ak977e/
ak977e00.pdf.
Kosuri, S., Eroshenko, N., Leproust, E.M., Super, M., Way, J., Li, J.B., and
Church, G.M. (2010). Scalable gene synthesis by selective amplification of
DNA pools from high-fidelity microchips. Nat. Biotechnol. 28, 1295–1299.
Friedland, A.E., Lu, T.K., Wang, X., Shi, D., Church, G., and Collins, J.J. (2009).
Synthetic gene networks that count. Science 324, 1199–1202.
Kosuri, S., Goodman, D.B., Cambray, G., Mutalik, V.K., Gao, Y., Arkin, A.P.,
Endy, D., and Church, G.M. (2013). Composability of regulatory sequences
controlling transcription and translation in Escherichia coli. Proc. Natl. Acad.
Sci. USA 110, 14024–14029.
Gaj, T., Gersbach, C.A., and Barbas, C.F., 3rd. (2013). ZFN, TALEN, and
CRISPR/Cas-based methods for genome engineering. Trends Biotechnol.
31, 397–405.
Gardner, T.S., and Hawkins, K. (2013). Synthetic biology: evolution or revolution? A co-founder’s perspective. Curr. Opin. Chem. Biol. 17, 871–877.
Gardner, T.S., Cantor, C.R., and Collins, J.J. (2000). Construction of a genetic
toggle switch in Escherichia coli. Nature 403, 339–342.
Garg, A., Lohmueller, J.J., Silver, P.A., and Armel, T.Z. (2012). Engineering
synthetic TAL effectors with orthogonal target sites. Nucleic Acids Res. 40,
7584–7595.
Gibson, D.G., Young, L., Chuang, R.Y., Venter, J.C., Hutchison, C.A., 3rd, and
Smith, H.O. (2009). Enzymatic assembly of DNA molecules up to several hundred kilobases. Nat. Methods 6, 343–345.
Gibson, D.G., Glass, J.I., Lartigue, C., Noskov, V.N., Chuang, R.Y., Algire,
M.A., Benders, G.A., Montague, M.G., Ma, L., Moodie, M.M., et al. (2010).
Creation of a bacterial cell controlled by a chemically synthesized genome.
Science 329, 52–56.
Gilbert, W. (1978). Why genes in pieces? Nature 271, 501.
Glass, J.I., Assad-Garcia, N., Alperovich, N., Yooseph, S., Lewis, M.R., Maruf,
M., Hutchison, C.A., 3rd, Smith, H.O., and Venter, J.C. (2006). Essential genes
of a minimal bacterium. Proc. Natl. Acad. Sci. USA 103, 425–430.
Kung, Y., Runguphan, W., and Keasling, J.D. (2012). From fields to fuels:
recent advances in the microbial production of biofuels. ACS Synth. Biol. 1,
498–513.
Lajoie, M.J., Rovner, A.J., Goodman, D.B., Aerni, H.R., Haimovich, A.D., Kuznetsov, G., Mercer, J.A., Wang, H.H., Carr, P.A., Mosberg, J.A., et al. (2013).
Genomically recoded organisms expand biological functions. Science 342,
357–360.
Lienert, F., Torella, J.P., Chen, J.H., Norsworthy, M., Richardson, R.R., and
Silver, P.A. (2013). Two- and three-input TALE-based AND logic computation
in embryonic stem cells. Nucleic Acids Res. 41, 9967–9975.
Mathews, C.K. (1993). The cell-bag of enzymes or network of channels?
J. Bacteriol. 175, 6377–6381.
Meister, G.E., and Joshi, N.S. (2013). An engineered calmodulin-based
allosteric switch for Peptide biosensing. ChemBioChem 14, 1460–1467.
Merrihew, R.V., Marburger, K., Pennington, S.L., Roth, D.B., and Wilson, J.H.
(1996). High-frequency illegitimate integration of transfected DNA at preintegrated target sites in a mammalian genome. Mol. Cell. Biol. 16, 10–18.
Mestas, J., and Hughes, C.C. (2004). Of mice and not men: differences
between mouse and human immunology. J. Immunol. 172, 2731–2738.
Goodman, D.B., Church, G.M., and Kosuri, S. (2013). Causes and effects of
N-terminal codon bias in bacterial genes. Science 342, 475–479.
Moon, T.S., Yoon, S.H., Lanza, A.M., Roy-Mayhew, J.D., and Prather, K.L.
(2009). Production of glucaric acid from a synthetic pathway in recombinant
Escherichia coli. Appl. Environ. Microbiol. 75, 589–595.
Gross, G., Waks, T., and Eshhar, Z. (1989). Expression of immunoglobulin-Tcell receptor chimeric molecules as functional receptors with antibody-type
specificity. Proc. Natl. Acad. Sci. USA 86, 10024–10028.
Murnane, J.P., Yezzi, M.J., and Young, B.R. (1990). Recombination events
during integration of transfected DNA into normal human cells. Nucleic Acids
Res. 18, 2733–2738.
Guntas, G., Mansell, T.J., Kim, J.R., and Ostermeier, M. (2005). Directed evolution of protein switches and their application to the creation of ligand-binding
proteins. Proc. Natl. Acad. Sci. USA 102, 11224–11229.
Mutalik, V.K., Guimaraes, J.C., Cambray, G., Mai, Q.A., Christoffersen, M.J.,
Martin, L., Yu, A., Lam, C., Rodriguez, C., Bennett, G., et al. (2013a). Quantitative estimation of activity and quality for collections of functional genetic
elements. Nat. Methods 10, 347–353.
Guye, P., Li, Y., Wroblewska, L., Duportet, X., and Weiss, R. (2013). Rapid,
modular and reliable construction of complex mammalian gene circuits.
Nucleic Acids Res. 41, e156.
Hutchison, C.A., Peterson, S.N., Gill, S.R., Cline, R.T., White, O., Fraser, C.M.,
Smith, H.O., and Venter, J.C. (1999). Global transposon mutagenesis and
a minimal Mycoplasma genome. Science 286, 2165–2169.
Huynen, M.A., and Bork, P. (1998). Measuring genome evolution. Proc. Natl.
Acad. Sci. USA 95, 5849–5856.
Isaacs, F.J., Dwyer, D.J., Ding, C., Pervouchine, D.D., Cantor, C.R., and
Collins, J.J. (2004). Engineered riboregulators enable post-transcriptional
control of gene expression. Nat. Biotechnol. 22, 841–847.
Kahl, L.J., and Endy, D. (2013). A survey of enabling technologies in synthetic
biology. J. Biol. Eng. 7, 13.
Keasling, J.D. (2012). Synthetic biology and the development of tools for metabolic engineering. Metab. Eng. 14, 189–195.
Keasling, J.D., Mendoza, A., and Baran, P.S. (2012). Synthesis: A constructive
debate. Nature 492, 188–189.
Khalil, A.S., Lu, T.K., Bashor, C.J., Ramirez, C.L., Pyenson, N.C., Joung, J.K.,
and Collins, J.J. (2012). A synthetic biology framework for programming
eukaryotic transcription functions. Cell 150, 647–658.
Knight, T.F. (2003). Idempotent vector design for standard assembly of
BioBricks. MIT Synthetic Biology Working Group Technical Reports. ([http://
hdl.handle.net/1721.1/21168]).
Kobayashi, H., Kaern, M., Araki, M., Chung, K., Gardner, T.S., Cantor, C.R.,
and Collins, J.J. (2004). Programmable cells: interfacing natural and engineered gene networks. Proc. Natl. Acad. Sci. USA 101, 8414–8419.
160 Cell 157, March 27, 2014 ª2014 Elsevier Inc.
Mutalik, V.K., Guimaraes, J.C., Cambray, G., Lam, C., Christoffersen, M.J.,
Mai, Q.A., Tran, A.B., Paull, M., Keasling, J.D., Arkin, A.P., and Endy, D.
(2013b). Precise and reliable gene expression via standard transcription
and translation initiation elements. Nat. Methods 10, 354–360.
Oberhardt, M.A., Palsson, B.O., and Papin, J.A. (2009). Applications of
genome-scale metabolic reconstructions. Mol. Syst. Biol. 5, 320.
Porter, D.L., Levine, B.L., Kalos, M., Bagg, A., and June, C.H. (2011). Chimeric
antigen receptor-modified T cells in chronic lymphoid leukemia. N. Engl. J.
Med. 365, 725–733.
Rader, R.A. (2012). FDA biopharmaceutical product approvals and trends:
2011. http://www.biopharma.com/approvals_2011.html.
Ren, X., Tahimic, C.G., Katoh, M., Kurimasa, A., Inoue, T., and Oshimura, M.
(2006). Human artificial chromosome vectors meet stem cells: new prospects
for gene delivery. Stem Cell Rev. 2, 43–50.
Ribeiro de Almeida, C., Stadhouders, R., Thongjuea, S., Soler, E., and Hendriks, R.W. (2012). DNA-binding factor CTCF and long-range gene interactions
in V(D)J recombination and oncogene activation. Blood 119, 6209–6218.
Robinson-Mosher, A., Shinar, T., Silver, P.A., and Way, J. (2013). Dynamics
simulations for engineering macromolecular interactions. Chaos 23, 025110.
Salis, H.M. (2011). The ribosome binding site calculator. Methods Enzymol.
498, 19–42.
Schirmer, A., Rude, M.A., Li, X., Popova, E., and del Cardayre, S.B. (2010).
Microbial biosynthesis of alkanes. Science 329, 559–562.
Sharma, S.S., Blattner, F.R., and Harcum, S.W. (2007). Recombinant protein
production in an Escherichia coli reduced genome strain. Metab. Eng. 9,
133–141.
Shetty, R.P., Endy, D., and Knight, T.F., Jr. (2008). Engineering BioBrick
vectors from BioBrick parts. J. Biol. Eng. 14, 5.
Silhavy, T.J., Casadaban, M.J., Shuman, H.A., and Beckwith, J.R. (1976).
Conversion of beta-galactosidase to a membrane-bound state by gene fusion.
Proc. Natl. Acad. Sci. USA 73, 3423–3427.
Siuti, P., Yazbek, J., and Lu, T.K. (2013). Synthetic circuits integrating logic and
memory in living cells. Nat. Biotechnol. 31, 448–452.
Steen, E.J., Kang, Y., Bokinsky, G., Hu, Z., Schirmer, A., McClure, A., Del Cardayre, S.B., and Keasling, J.D. (2010). Microbial production of fatty-acidderived fuels and chemicals from plant biomass. Nature 463, 559–562.
Tinberg, C.E., Khare, S.D., Dou, J., Doyle, L., Nelson, J.W., Schena, A., Jankowski, W., Kalodimos, C.G., Johnsson, K., Stoddard, B.L., and Baker, D.
(2013). Computational design of ligand-binding proteins with high affinity
and selectivity. Nature 501, 212–216.
Wang, H.H., Isaacs, F.J., Carr, P.A., Sun, Z.Z., Xu, G., Forest, C.R., and
Church, G.M. (2009). Programming cells by multiplex genome engineering
and accelerated evolution. Nature 460, 894–898.
Wei, K.Y., Chen, Y.Y., and Smolke, C.D. (2013). A yeast-based rapid prototype
platform for gene control elements in mammalian cells. Biotechnol. Bioeng.
110, 1201–1210.
Whitaker, W.R., and Dueber, J.E. (2011). Metabolic pathway flux enhancement
by synthetic protein scaffolding. Methods Enzymol. 497, 447–468.
Wieland, M., and Fussenegger, M. (2012). Engineering molecular circuits
using synthetic biology in mammalian cells. Annu. Rev. Chem. Biomol. Eng.
3, 209–234.
Win, M.N., Liang, J.C., and Smolke, C.D. (2009). Frameworks for programming biological function through RNA parts and devices. Chem. Biol. 16,
298–310.
Yeh, B.J., Rutigliano, R.J., Deb, A., Bar-Sagi, D., and Lim, W.A. (2007). Rewiring cellular morphology pathways with synthetic guanine nucleotide exchange
factors. Nature 447, 596–600.
Young, R., and Bremer, H. (1976). Polypeptide-chain-elongation rate
in Escherichia coli B/r as a function of growth rate. Biochem. J. 160,
185–194.
Cell 157, March 27, 2014 ª2014 Elsevier Inc. 161